A general spatio-temporal model for environmental data

The modern statistical models for spatio-temporal environmental data are being increasingly used and are particularly useful, for example, for epidemiological studies, dynamical risk mapping and sensitivity analysis. In this paper, we consider a hierarchical spatio-temporal model estimated by a combination of the EM algorithm and bootstrap simulations. To do this, due to the high computational load required, we use an appropriate distributed environment. The method is illustrated by an application on air pollution data which considers model, standard errors and confidence intervals estimation. The computational load is managed by parallel computing techniques implemented on a computer cluster using R software. The algorithm performance is analyzed in terms of the convergence iterations and computing times.